You are here

  • Home
  • Archive
  • Volume 21, Issue 2
  • Do nurse and patient injuries share common antecedents? An analysis of associations with safety climate and working conditions

Article Text

Article menu
Original research
Do nurse and patient injuries share common antecedents? An analysis of associations with safety climate and working conditions
  1. Jennifer A Taylor1,
  2. Francesca Dominici2,
  3. Jacqueline Agnew3,
  4. Daniel Gerwin1,
  5. Laura Morlock4,
  6. Marlene R Miller4,5
  1. 1Department of Environmental and Occupational Health, Drexel University School of Public Health, Philadelphia, Pennsylvania, USA
  2. 2Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts, USA
  3. 3Department of Environmental Health Sciences, Division of Occupational and Environmental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  4. 4Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  5. 5Department of Pediatrics, Johns Hopkins Children's Center, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  1. Correspondence to Jennifer A Taylor, Assistant Professor, Department of Environmental and Occupational Health, Drexel University School of Public Health, 1505 Race Street, MS 1034, Philadelphia, PA 19102; jat65{at}drexel.edu

Abstract

Background Safety climate and nurses' working conditions may have an impact on both patient outcomes and nurse occupational health, but these outcomes have rarely been examined concurrently.

Objective To examine the association of unit-level safety climate and specific nurse working conditions with injury outcomes for both nurses and patients in a single hospital.

Research design A cross-sectional study was conducted using nursing-unit level and individual-level data at an urban, level-one trauma centre in the USA. Multilevel logistic regressions were used to examine associations among injury outcomes, safety climate and working conditions on 29 nursing units, including a total of 723 nurses and 28 876 discharges.

Measures Safety climate was measured in 2004 using the Safety Attitudes Questionnaire (SAQ). Working conditions included registered nursing hours per patient day (RNHPPD) and unit turnover. Patient injuries included 290 falls, 167 pulmonary embolism/deep vein thrombosis (PE/DVT), and 105 decubitus ulcers. Nurse injury was defined as a reported needle-stick, splash, slip, trip, or fall (n=78). Working conditions and outcomes were measured in 2005.

Results The study found a negative association between two SAQ domains, Safety and Teamwork, with the odds of both decubitus ulcers and nurse injury. RNHPPD showed a negative association with patient falls and decubitus ulcers. Unit turnover was positively associated with nurse injury and PE/DVT, but negatively associated with falls and decubitus ulcers.

Conclusions Safety climate was associated with both patient and nurse injuries, suggesting that patient and nurse safety may actually be linked outcomes. The findings also indicate that increased unit turnover should be considered a risk factor for nurse and patient injuries.

  • Organisational culture
  • injuries
  • medical errors
  • occupational safety
  • administrative nursing research
  • adverse events, epidemiology and detection
  • patient safety
  • safety culture
  • healthcare quality improvement
  • near miss

https://doi.org/10.1136/bmjqs-2011-000082

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Background

    Increasing attention is being paid to the impact of organisational safety climate and working conditions on healthcare outcomes.1–3 Safety climate refers to ‘shared employee perceptions of the priority of safety at their unit and organisation at large, especially in situations where safety competes with other performance facets such as speed of care or its quality’.2 Working conditions include an array of factors such as adequacy of resources, qualities of management and staffing characteristics.1 4 5

    Meaningful safety climate findings have emanated from investigating employee perceptions in single hospital studies because employee perceptions reflect their immediate work environment.6–8 Safety climate is local, demonstrating more variability within a hospital than between hospitals.8

    Meta-analytic studies have contributed to the understanding of safety climate's systemic role in healthcare processes and outcomes. These studies suggest the importance of organisational-level factors, including safety climate, on individual safety behaviour.9–12 Other previous research has explored relationships among safety climate and/or working conditions and nurse occupational injuries,5 6 13 patient injuries,4 14 15 and nurse and patient injuries together.16 A detailed summary of these studies is presented in supplementary online appendix 1. The study by Hofmann and Mark found that a unit's safety climate is an important predictor of nurse occupational safety (back injuries) as well as patient outcomes (medication errors, urinary tract infections) and was the only study to include both nurse and patient outcomes.16

    This study extends the literature by examining a single hospital to investigate the extent to which the same organisational characteristics (working conditions and safety climate) predict injuries for both patients and the nurses who care for them, an important question given that nurses constitute the largest workforce in healthcare.17

    Methods

    Study design

    The study was cross-sectional with a longitudinal aspect in that safety climate data were assessed in 2004 while injury outcomes were collected in 2005. This study investigated whether unit-level working conditions and safety climate were predictors of both individual-level nurse and patient injuries. All study activities were approved by the investigators' institutional review board. Health Insurance Portability and Accountability Act approval was given by the hospital.

    This study was conducted at a large, urban, level-one trauma centre with Magnet nursing status in the USA.18 In 2005, the hospital had approximately 1900 total beds and 48 000 discharges. To be included, nursing units had to have a 60% or greater safety climate survey response rate among registered nurses, available turnover data, and information on hours of direct nursing care. Twenty-nine of 46 (63%) direct care inpatient nursing units satisfied these criteria. The 29 units included 19 medical, five surgical, four medical/surgical and one rehabilitation unit. These units encompassed paediatrics, general medicine, psychiatry, neurology, oncology, gynaecology and general surgery. The smallest of these units had a total of three nurses on staff and the largest 65, with a mean of 25 nurses on staff per unit. Multiple attempts were made to attain a 60% response rate on each unit, and the survey designer felt it critical to exclude units with less than 60% response rate. Among the 29 units included in this study, the response rate varied from 60% to 100%, with an average response rate of 76%.

    The 60% response rate threshold is of particular value in a safety climate study because research on non-response bias has found that non-respondents show lower organisational commitment and job satisfaction, and greater intention to quit.19 Such findings suggest that Safety Attitudes Questionnaire (SAQ) non-respondents may have different safety perceptions than respondents, and a higher response rate threshold reduces the risk of non-response bias. A recent meta-analysis of response rates in 2037 studies conducted from 1995 to 2008 found that for personally distributed surveys of non-managerial employees (as with the nurses in this study) average response rates are fairly high, with a 69% response rate at the 50th percentile of all surveys analysed.20 This finding further supports the appropriateness of the 60% minimum response rate in the study design, as well as the utility of the 76% average response rate among included units.

    Explanatory variables

    Safety climate: the safety attitudes questionnaire

    The SAQ elicits healthcare workers' safety climate perceptions. The SAQ has been described previously.21–23 The hospital administered the SAQ to its employees in 2004, and the Department of Risk Management provided the resultant data. The 36-item survey uses a five-point Likert scale to elicit staff attitudes. Responses to questions within each domain are averaged and converted to a 100-point scale. A high score indicates greater agreement that a positive safety climate exists on a given unit. All average domain scores were rescaled (divide by a factor of 10) and interpreted as a 10-point change in the average domain score. The SAQ domains and questions are presented in supplementary online appendix 2.

    Working conditions: unit turnover and registered nursing hours per patient day

    The unit-turnover rate was defined as the total number of registered nurse full-time equivalents (FTEs) who voluntarily left one unit for another in 1 year, divided by the number of registered nurse FTEs assigned to the unit in that year. Measuring FTEs includes and controls for the proportionate contributions of part-time and full-time employees. Termination (discontinuation of employment either voluntarily or involuntarily) was not included in this study because the hospital's director of nursing felt that termination more strongly represents a nurse's overall experience with the hospital rather than with the nurse's particular unit. The turnover rate was rescaled (multiplied by a factor of 10) and interpreted as a 10% change in the rate to make result interpretation more applicable to hospitals' actual goal setting. The Department of Human Resources provided 2005 data on nurse turnover for the numerator of the unit-turnover metric, and total FTE nurses for each nursing unit in 2005, establishing the denominator.

    Registered nursing hours per patient day (RNHPPD), as defined by the American Nurses Association (ANA)/National Quality Forum (NQF), is the annual sum of agency and hospital registered nurse (RN) productive hours (direct patient care activities as opposed to administrative or other tasks) divided by the total number of patient days in the calendar year, as follows:

    RNHPPD=productive agency RN hours+productive hospital RN hourstotal number of patient days/year

    Nursing hours and total patient days for 2005 were acquired from the hospital's Department of Nursing. Average RNHPPD values were calculated for each nursing unit.

    Outcome measures

    Patient injuries

    The nursing-sensitive patient injuries examined were patient falls, decubitus ulcers (‘pressure ulcers’), and pulmonary embolism/deep vein thrombosis (PE/DVT). These outcomes were chosen because of the availability of the data and because the ANA considers the first two injuries to be ‘nursing-sensitive indicators’, meaning they reflect the structure, process and outcomes of nursing care,24 and because all three injuries are considered preventable and may have severe repercussions. The Centers for Medicare and Medicaid Services no longer reimburses hospitals for costs associated with medical errors termed ‘never events’, occurring while in a healthcare facility, including patient death associated with a fall, stage 3 or 4 pressure ulcers acquired after admission, and patient death or serious disability associated with intravascular air embolism.25

    The Agency for Healthcare Research and Quality's publicly available Windows QI software, the Patient Safety Indicators (PSIs) (http://www.qualityindicators.ahrq.gov/psi_download.htm), was applied to the hospital's administrative discharge data to identify decubitus ulcers and PE/DVT. PSIs were used because the PSI algorithm accounts for patient diagnoses by adjusting the denominator to focus on patients for whom the relevant injury is most likely preventable.26 Patient falls were measured using data from Patient Safety Net (PSN) V.3.0, a voluntary, electronic reporting system for the occurrence of errors and near misses developed by the University HealthSystems Consortium (https://www.uhc.edu/11851.htm). All patient falls were selected regardless of whether injury resulted. PSN was chosen as the data source for falls because it provides the most comprehensive record of patient falls, as PSIs capture only those falls resulting in postoperative hip fracture. The hospital's Department of Risk Management provided patient-level outcomes from 2005. The categories for PSN and PSI are presented in supplementary online appendix 3.

    Nurse injuries

    Nurse injury was defined as the occurrence of a reported needle-stick, splash, slip, trip or fall. Occupational exposures to blood-borne pathogens through needle-sticks and splashes are of concern because of the risk and severe consequences of hepatitis B and C, and HIV and AIDS infection.27–29 Slips, trips and falls (STFs) are also important because their incidence rate in hospitals is much higher than that found in all other private industries. For example, Bell et al identified this magnitude to be 75% greater in a study in which the hospital incidence rate of STFs was 35.2 per 10 000 employees.30 Back injuries were excluded because the hospital felt that work relatedness was difficult to determine. Individual-level nurse injury data from 2005 were acquired from the Hospital's Division of Occupational Medicine. Nurse injuries were routinely recorded by the hospital in written injury reports which were then entered in a database.

    It is well established that increases in the number of patient comorbidities corresponds to increased length of stay, cost of care, and risk of dying in the hospital.31 32 These factors, in turn, can influence the patient outcomes of interest in this study. Accordingly, a patient complexity metric was included to control for the effect of each patient's comorbidities.

    Statistical analysis

    For each nursing unit in the study, unit turnover, RNHPPD, and the six domains of the SAQ constituted the main explanatory variables. For the patient outcome models, patient complexity was included as a control to address the potential confounding effect of the patient's condition. The All Patient Refined Diagnosis Related Group Patient complexity level was extracted from the hospital's discharge database. It is calculated using International Classification of Diseases, ninth revision (ICD9) diagnoses, ICD9 procedures, and patient age by the 3M APRDRG grouping software, V.12 (http://solutions.3m.com/wps/portal/3M/en_US/3M_Health_Information_Systems/HIS/Products/APRDRG_Software/). For the analysis, patient complexity was dichotomised into high (complexity level 3 or 4) or low (complexity level 1 or 2) categories to separate high and low severity of illness/mortality risk.

    STATA was used to perform all analyses (Stata 8, StataCorp LP, College Station, Texas, USA). Descriptive statistics were used to compare characteristics between nursing units with and without nurse injuries. Correlations among variables were explored using Pearson's product moment, and simple logistic regressions were conducted for each outcome to determine variable selection for the final model. To justify aggregation of SAQ domains at the nursing unit level, RWG(J) was calculated according to the method described by James et al.33 Coefficients of reliability (Cronbach's α) for each domain of the SAQ were calculated using the α command in Stata.

    The multilevel model preserves the structure of the data as it was collected by analysing outcomes at the individual level (patient and nurse injuries), while allowing explanatory covariates to be analysed at both individual levels (patient complexity) and group levels (unit working conditions, average unit SAQ scores). Intraclass correlation coefficients (ICCs) for each model were estimated according to Hox (2002):34ICC:ρ=σ2(σ2+π2/3)where σ2 is the between-unit variance and π2/3 is the within-unit variance expressed as a constant.

    For each regression, the unit of analysis was the patient (nested within the nursing unit) or the nurse (nested within the nursing unit). Individual outcomes on the same unit were presumed to be dependent. Multilevel logistic regression models with a random intercept were used to account for this clustering.

    The dependent variable in the nurse injury model was the log odds of the probability of individual-level nurse injury versus the probability of no injury. Seventy-eight nurses out of 723 experienced an injury. Unit turnover, RNHPPD, and the SAQ domains were included as unit-level explanatory covariates.

    The dependent variable in each patient injury model was the log odds of the probability of individual-level patient injury versus the probability of no injury. From the 28 260 discharges in the PSI dataset, 105 decubitus ulcers and 167 PE/DVT were identified. From the 28 876 discharges in the PSN dataset, 290 patient falls were identified. Unit turnover, RNHPPD and the SAQ domains were included as unit-level explanatory covariates. Patient complexity was included as a patient-level covariate.

    p Value statistics have been shown to be inappropriate for observational studies.35 36 ORs with 95% CIs are presented instead of p values to focus on the direction, magnitude and precision of the associations rather than hypothesis testing. While the results are highlighted for which the lower and upper bound of the CI are either above or below one, the results that slightly cross the null but whose magnitude and precision suggest that they are worthy of further consideration are also reported.

    Results

    Table 1 compares units with and without reported injuries, presenting unit-level summary statistics including total patient safety events and total nurse injuries. Units with patient falls had lower average SAQ scores than units without falls, and this trend was also evident for PE/DVT though less robust. No important differences were observed between the study variable means for the 29 units included and the 17 units excluded (data not shown).

    View this table:
    Table 1

    Characteristics of nursing units

    A Pearson correlation matrix for falls was run to elucidate correlations among variables for the final models. A representative table for falls is included as supplementary online appendix 4. The Pearson correlation matrix showed that scores for individual SAQ domains were highly correlated with one another (r=0.65–0.91), therefore each domain was examined in a separate model using the same panel of explanatory covariates. The SAQ Stress Recognition results were excluded due to concerns about this domain's construct validity, which requires additional examination beyond the scope of this study. The authors are preparing a separate analysis of this domain.

    Justification for aggregating individual responses to the SAQ by nursing unit is given in table 2. Sufficient within-unit homogeneity existed to warrant aggregation at the unit level (RWG(J) 0.643–0.831). Table 2 also presents Cronbach's α for each SAQ domain, which showed acceptable subscale reliability.

    View this table:
    Table 2

    Data aggregation results

    The ICC measures the degree of dependence in the outcome after accounting for the covariates. The larger the intraclass correlation, the more likely it is that nurse outcomes are correlated within their unit even after adjusting for the covariates. The ICCs for the random intercept logistic regressions were in the ‘medium’ to ‘large’ range.37 The effective sample size was calculated38 and it was found that the ICC did not significantly alter the effective sample size, thus preserving power (data not shown). Regardless of the ICC's size, the structure of the data and the study design still indicated the use of multilevel models as appropriate.

    Figure 1 displays a summary of the magnitude, precision, and direction of the effect estimates for the main explanatory covariates by each injury outcome. Table 3 shows the results of the multilevel models adjusted by patient complexity. In general, the domains of the SAQ showed negative associations with injury outcomes, though the degree of precision varied. The Safety and Teamwork models were the most convincing, with ORs under 1.0 and more precise CIs for nurse injury and decubitus ulcer. For each 10-point increase in the average SAQ Safety domain score, a 48% reduction was observed in the odds of decubitus ulcer (OR=0.52, 95% CI 0.29 to 0.92), as well as a 45% reduction in the odds of nurse injury (OR=0.55, 95% CI 0.32 to 0.97). The Teamwork domain behaved similarly: for each 10-point increase in the average Teamwork domain score, a 44% reduction in the odds of decubitus ulcer (OR=0.56, 95% CI 0.38 to 0.82) and a 40% reduction in the odds of nurse injury was found (OR=0.60, 95% CI 0.41 to 0.89). The Morale domain also showed a negative association with these outcomes, but slightly crossed the null.

    Figure 1
    Figure 1

    Comparison of 95% CIs by injury outcome and major explanatory covariates. Results for Safety, Morale, and Teamwork are from their respective models (see table 3). Registered nursing hours per patient day (RNHPPD) and Turnover results are from the Teamwork model.

    View this table:
    Table 3

    Results of multilevel logistic regression with random intercept: injury type by SAQ domain

    In many of the models, working conditions were associated with both patient and nurse injury. As expected, RNHPPD showed a negative association with falls and decubitus ulcers, with falls having the more precise estimates. Controlling for all other covariates, in the Safety model each additional hour of RNHPPD was associated with a 9% decrease in the odds of patient falls (OR=0.91, 95% CI 0.83 to 1.00). RNHPPD slightly exacerbated the risk of PE/DVT, an effect of reasonable precision, and showed no effect on nurse injury. Unit turnover increased the risk of nurse injury and PE/DVT: in the Teamwork model, unit turnover was positively associated with nurse injury (OR=1.68, 95% CI 1.00 to 2.84) and with PE/DVT (OR=1.15, 95% CI 0.97 to 1.37). These estimates appear to be meaningful given the consistent direction of the intervals away from the null, and the narrow CIs. In the Teamwork model, higher unit turnover reduced the odds for falls (OR=0.84, 95% CI 0.69 to 1.02).

    The estimates for RNHRPPD and unit turnover were virtually the same in all models, and the results from the Teamwork model are discussed as exemplars. Either Teamwork or Safety could have been chosen because they produced the most precise results among the models. Among the SAQ domains, Teamwork and Safety are the best understood (Sexton, personal communication, 2007) because interventions targeting both the Teamwork and Safety domains have already been developed and tested, with impressive results.39 40

    Discussion

    This study explored associations of safety climate and nurse working conditions with injury outcomes for both nurses and patients. Meaningful relationships were found between the safety climate domains of Safety and Teamwork and decubitus ulcer and nurse injury, between nursing hours per patient day (RNHPPD) and patient falls, and between unit turnover and both patient and nurse injuries. These findings suggest that nurse and patient injuries share common underlying causes.

    A negative association was found between two SAQ domains, Safety (perception of organisational commitment to safety) and Teamwork (perception of the quality of collaboration), with the odds of both decubitus ulcer and nurse injury. For each 10-point increase in the average Safety or Teamwork score, the odds of decubitus ulcer declined by 44–48% and the odds of nurse injury by 40–45%. These associations align with the findings of a recent meta-analysis of 203 independent samples in which social support, leadership, and safety climate were all key aspects of a supportive environment which explained the most variance in safety outcomes across industries.9

    Given that decubitus ulcers are most common among relatively immobile patients, these findings raise interesting questions for care delivery. Nursing care for immobilised patients is generally more exhausting because regular lifting is required. The demands of caring for such patients could increase the importance of teamwork and have a special impact on safety. It thus seems reasonable that the Teamwork and Safety domains could reflect the potential for fatigue and other injury risk factors associated with caring for this patient population, as well as these patients' risk of decubitus ulcer.

    The finding of a negative relationship between RNHPPD and patient falls corroborates previous findings of Krauss et al, who demonstrated that patient falls are sensitive to nurse-to-patient ratios, with patients seven times more likely to fall when the nurse was assigned seven or more patients than when the nurse had three patients or fewer.41 The advantage of increased nursing time for patients at risk for falls seems intuitive: more patient contact creates additional capacity for direct observation, and thus more opportunities to prevent a fall. The slightly exacerbated risk of PE/DVT with increasing RNHPPD is counter intuitive and deserves more careful study using a longitudinal study design.

    A positive association was found between the odds of PE/DVT and the unit-turnover rate, but a negative association between unit turnover and the rates of both patient falls and decubitus ulcers. Increased nurse turnover has been associated with longer patient length of stay and higher costs per discharge,42 but knowledge is still emerging regarding the relationship between nurse turnover and patient outcomes.43–46 A recent study by Bae et al used a unit-turnover metric identical to the one used in this study, and found that higher rates of unit turnover were associated with lower levels of patient falls, consistent with our finding.47 Similarly and equally surprising, Mark et al found that units with more experienced nurses had both lower turnover and higher fall rates.48

    Further, a 68% increase in the odds of nurse injury was found with each 10% increase in the unit-turnover rate. While evidence links organisational climate and nurse turnover,5 49 50 little has been established about the direct effect of turnover on occupational safety outcomes, though such research has been recommended.43 44 46 Higher unit-turnover rates may result in a greater proportion of nurses who are new to the unit, and therefore less experienced with the unit's specific procedures, techniques and equipment, putting nurses at greater risk of injury. A study of worker's compensation records found that significant changes in job tasks or title may essentially constitute a ‘new job’, and also found an inverse relationship between job tenure and claim rates, with workers in the first month of a job four times more likely to have a lost-time claim than those with over 1 year on the job.51 In the nursing context specifically, it has been found that nurses with less than 5 years of experience are at greater risk of sharps injuries.29 In a related finding among home healthcare workers, the risk of sharps injuries and blood/body fluid exposures increases across those holding full-time, part time and per-diem employment.52

    Unit turnover, in sum, was positively associated with nurse injury and the odds of PE/DVT, but negatively associated with the odds of patient falls and decubitus ulcers. These relationships indicate that while unit turnover is a common predictor for both nurse and patient injuries, the causal connections are likely to be complex. Further study is warranted to explore the specific processes that might explain the differing associations between unit turnover and each of these safety outcomes.

    The results show the importance of controlling for patient complexity when investigating patient injury outcomes. Patient complexity shows a strong positive association with each of the patient injury outcomes investigated. This finding is consistent with the literature on associations between patient complexity and adverse outcomes.31

    The authors cannot determine whether a nurse injury preceded unit turnover or whether unit turnover was a causal factor for nurse injury due to the cross-sectional aspects of this study. However, the study design is strengthened by the fact that nurse safety perceptions were measured before the safety outcome data were collected. A meta-analysis by Beus et al demonstrates that injuries have a significant predictive effect on safety climate, especially when measured at the organisational level, necessitating a prospective measurement of safety climate by researchers investigating safety climate's effect on injuries.12 Under-reporting of nurse injury is also a concern because the data are based on voluntary injury reports, and it is well established that occupational injuries are under-reported by healthcare workers.53 54 On balance, the authors believe that the associations identified are meaningful and worthy of further investigation using other methods, including classic longitudinal designs. Regarding the restriction of this study to the 29 units with at least a 60% response rate, it is conceivable that low response rate units were different from those that were included. The survey included secondary data available to the investigator only after administration by the hospital. To the investigators' knowledge, no effort was made to gather additional data about low response rate units. However, the low response rate units did not have significantly different means for the study variables under investigation (data not shown). While the single-hospital setting and the restriction to the higher-response units limit the generalisability of this study, the associations identified can now be further explored in a multi-hospital study.

    In addition, unit type or skill mix was not investigated. Lake et al found variability in patient falls by unit type.55 Moreover, Dunton et al have shown that in describing the negative association of RNHPPD with patient falls, the magnitude of the effect size varies by unit type.56 The control of patient complexity in this study may address some of this variability, but does not do so entirely. A skill mix metric would help identify the proportion of licensed versus unlicensed nursing personnel, which in combination with higher RNHPPD is generally associated with decreased patient injuries.57 58 Unruh suggests that there is a threshold skill mix level beyond which an increase in the proportion of licensed nurses has no additional influence.58 Since RNHPPD was included in this study, it seems unlikely that adding skill mix to the model would have substantially changed the results.

    The unit-turnover metric accounted for nurses who left one unit for another, but did not account for nurses who left the hospital entirely. The latter group is captured by commonly used termination metrics. In addition to the conceptual reasons for excluding this variable from the study (the global nature of termination vs the local nature of unit turnover), exploratory data analysis found that termination was not significant in the bivariate results for either nurse or patient injury (data not shown). The authors expect the inclusion of a combined metric for unit turnover and termination would not have changed the direction of the observed associations, but may have increased the magnitude of the overall effect. These two metrics offer insight into different aspects of nurses' experience and may fruitfully be used together in future research.

    It is important to note that the science underlying the sensitivity of the SAQ is still in its infancy. The SAQ was used in this study because of the available surveys measuring hospital safety climate, only the SAQ had been used to explore the relationship between safety climate scores, patient outcomes and nurse turnover.59 Sexton reported that increasingly positive perceptions on four of the six SAQ domains were associated with shorter intensive care unit stays, lower medication errors and decreased nurse turnover.22 23 60 As further research explores the SAQ's sensitivity compared with actual changes in the healthcare environment, it will be possible to draw increasingly precise conclusions from studies involving this instrument.

    Conclusion

    This study demonstrates that nurses and patients share similar organisational risk factors for injury. The strength of evidence is most convincing from the Teamwork and Safety domains of the SAQ, each of which was negatively associated with decubitus ulcers and nurse injuries. This study also demonstrates that unit turnover should be analysed not merely as an organisational outcome, but should also be among the working conditions considered as a risk factor for nurse and patient injuries. These results suggest that patient and nurse safety may constitute linked outcomes rather than distinct silos. Traditionally, the fields of occupational and patient safety have been addressed separately, but regarding them as related components of an organisation's safety culture might be more apt. Future healthcare safety research and intervention efforts should consider a new paradigm integrating lessons learnt from both disciplines to more comprehensively and efficiently reduce preventable injuries to both workers and patients.

    Acknowledgments

    The authors wish to acknowledge the anonymous reviewers for their thoughtful contributions to the revision of this manuscript. In addition, the authors express their deep gratitude to J. Bryon Sexton, PhD, MA, the developer of the SAQ, for his advice and encouragement in their application of the survey. In addition, for their data support and technical assistance, the authors would like to acknowledge: Patty Dawson, MSN, RN, Magnet Program and Clinical Outcomes Coordinator, Johns Hopkins Hospital; Karen Haller RN, PhD, FAAN, Vice President for Nursing and Patient Care Services, Johns Hopkins Hospital; and Lori Paine, RN, MS, Director, Patient Safety, Johns Hopkins Hospital and Center for Innovation in Quality Patient Care. Special thanks to Igor Burstyn, PhD, for his meaningful statistical insights and Ravi Pandian, MA, Drexel University School of Public Health, for his assistance with homogeneity coefficient statistics.

    References

    1. ↵
      1. Lundstrom T,
      2. Pugliese G,
      3. Bartley J,
      4. et al
      . Organizational and environmental factors that affect worker health and safety and patient outcomes. Am J Infect Control 2002;30:93–106.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Zohar D,
      2. Livne Y,
      3. Tenne-Gazit O,
      4. et al
      . Healthcare climate: a framework for measuring and improving patient safety. Crit Care Med 2007;35:1312–17.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Mannion R,
      2. Konteh FH,
      3. Davies HT
      . Assessing organisational culture for quality and safety improvement: a national survey of tools and tool use. Qual Saf Health Care 2009;18:153–6.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Stone PW,
      2. Mooney-Kane C,
      3. Larson EL,
      4. et al
      . Nurse working conditions and patient safety outcomes. Med Care 2007;45:571–8.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Stone PW,
      2. Gershon RR
      . Nurse work environments and occupational safety in intensive care units. Policy Polit Nurs Pract 2006;7:240–7.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Gershon RRM,
      2. Karkashian CD,
      3. Grosch JW,
      4. et al
      . Hospital safety climate and its relationship with safe work practices and workplace exposure incidents. Am J Infect Control 2000;28:211–21.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Neal A,
      2. Griffin MA,
      3. Hart PM
      . The impact of organizational climate on safety climate and individual behavior. Saf Sci 2000;34:99–109.
      OpenUrlCrossRefWeb of Science
    8. ↵
      1. Pronovost PJ,
      2. Weast B,
      3. Holzmueller CG,
      4. et al
      . Evaluation of the culture of safety: survey of clinicians and managers in an academic medical center. Qual Saf Health Care 2003;12:405–10.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Nahrgang JD,
      2. Morgeson FP,
      3. Hofmann DA
      . Safety at work: a meta-analytic investigation of the link between job demands, job resources, burnout, engagement, and safety outcomes. J Appl Psychol 2011;96:71–94.
      OpenUrlCrossRefPubMed
    10. ↵
      1. Christian MS,
      2. Bradley JC,
      3. Wallace JC,
      4. et al
      . Workplace safety: a meta-analysis of the roles of person and situation factors. J Appl Psychol 2009;94:1103–27.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Clarke S
      . An integrative model of safety climate: linking psychological climate and work attitudes to individual safety outcomes using meta analysis. J Occup Organ Psychol 2010;83:553–78.
      OpenUrlCrossRefWeb of Science
    12. ↵
      1. Beus JM,
      2. Payne SC,
      3. Bergman ME,
      4. et al
      . Safety climate and injuries: an examination of theoretical and empirical relationships. J Appl Psychol 2010;95:713–27.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Mark BA,
      2. Hughes LC,
      3. Belyea M,
      4. et al
      . Does safety climate moderate the influence of staffing adequacy and work conditions on nurse injuries? J Safety Res 2007;38:431–46.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Felknor SA,
      2. Aday LA,
      3. Burau KD,
      4. et al
      . Safety climate and its association with injuries and safety practices in public hospitals in Costa Rica. Int J Occup Environ Health 2000;6:18–25.
      OpenUrlPubMed
    15. ↵
      1. Katz-Navon TAL,
      2. Naveh E,
      3. Stern Z
      . Safety climate in health care organizations: a multidimensional approach. Acad Manage J 2005;48:1075.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Hofmann DA,
      2. Mark B
      . An investigation of the relationship between safety climate and medication errors as well as other nurse and patient outcomes. Person Psychol 2006;59:847–69.
      OpenUrlCrossRef
    17. ↵
      1. Stone PW,
      2. Du Y,
      3. Gershon RR
      . Organizational climate and occupational health outcomes in hospital nurses. J Occup Environ Med 2007;49:50–8.
      OpenUrlCrossRefPubMed
    18. ↵
      American Nurses Credentialing Center. ANCC Magnet Recognition Program. American Nurses Credentialing Center, 2010. http://www.nursecredentialing.org/Magnet.aspx (accessed Mar 2011).
    19. ↵
      1. Rogelberg SG,
      2. Luong A,
      3. Sederburg ME,
      4. et al
      . Employee attitude surveys: examining the attitudes of noncompliant employees. J Appl Psychol 2000;85:284–93.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Anseel F,
      2. Lievens F,
      3. Schollaert E,
      4. et al
      . Response rates in organizational science, 1995–2008: a meta-analytic review and guidelines for survey researchers. J Bus Psychol 2010;25:335–49.
      OpenUrlCrossRef
    21. ↵
      1. Sexton JB,
      2. Thomas EJ,
      3. Helmreich RL
      . Error, stress, and teamwork in medicine and aviation: cross sectional surveys. BMJ 2000;320:745–9.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Sexton JB,
      2. Holzmueller CG,
      3. Pronovost PJ,
      4. et al
      . Variation in caregiver perceptions of teamwork climate in labor and delivery units. J Perinatol 2006;26:463–70.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Sexton JB
      . A Matter of Life or Death: Social Psychological and Organizational Factors Related to Patient Outcomes in the Intensive Care Unit. Unpublished PhD dissertation University of Texas at Austin, 2002.
    24. ↵
    25. ↵
      Centers for Medicare and Medicaid Services. Eliminating Serious, Preventable, and Costly Medical Errors—Never Events. Centers for Medicare and Medicaid Services, 2006. http://www.cms.gov/apps/media/press/release.asp?Counter=1863 (accessed 13 Jun 2011).
    26. ↵
      1. Rosen AK,
      2. Rivard P,
      3. Zhao S,
      4. et al
      . Evaluating the patient safety indicators: how well do they perform on Veterans Health Administration data? Medical Care 2005;43:873–84.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Clarke SP,
      2. Sloane DM,
      3. Aiken LH
      . Effects of hospital staffing and organizational climate on needlestick injuries to nurses. Am J Public Health 2002;92:1115–19.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Clarke SP,
      2. Rockett JL,
      3. Sloane DM,
      4. et al
      . Organizational climate, staffing, and safety equipment as predictors of needlestick injuries and near-misses in hospital nurses. Am J Infect Control 2002;30:207–16.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Clarke SP
      . Hospital work environments, nurse characteristics, and sharps injuries. Am J Infect Control 2007;35:302–9.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Bell JL,
      2. Collins JW,
      3. Wolf L,
      4. et al
      . Evaluation of a comprehensive slip, trip and fall prevention programme for hospital employees. Ergonomics 2008;51:1906–25.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Elixhauser A,
      2. Steiner C,
      3. Harris DR,
      4. et al
      . Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. Song X,
      2. Srinivasan A,
      3. Plaut D,
      4. et al
      . Effect of nosocomial vancomycin-resistant enterococcal bacteremia on mortality, length of stay, and costs. Infect Control Hosp Epidemiol 2003;24:251–6.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. James LR,
      2. Demaree RG,
      3. Wolf G
      . Estimating within-group interrater reliability with and without response bias. J Appl Psychol 1984;69:85.
      OpenUrlCrossRefWeb of Science
    34. ↵
      1. Hox J
      . Multilevel Analysis: Techniques and Applications. Mahwah, NJ: Lawrence Erlbaum Associates, 2002.
    35. ↵
      1. Poole C
      . Low P-values or narrow confidence intervals: which are more durable? Epidemiology 2001;12:291–4.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Greenland S
      . Randomization, statistics, and causal inference. Epidemiology 1990;1:421–9.
      OpenUrlCrossRefPubMed
    37. ↵
      1. Zyzanski SJ,
      2. Flocke SA,
      3. Dickinson LM
      . On the nature and analysis of clustered data. Ann Fam Med 2004;2:199–200.
      OpenUrlFREE Full Text
    38. ↵
      1. Killip S,
      2. Mahfoud Z,
      3. Pearce K
      . What is an intracluster correlation coefficient? Crucial concepts for primary care researchers. Ann Fam Med 2004;2:204–8.
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Pronovost P,
      2. Sexton B
      . Assessing safety culture: guidelines and recommendations. Qual Safety Health Care 2005;14:231–3.
      OpenUrlCrossRefWeb of Science
    40. ↵
      1. Thomas EJ,
      2. Sexton JB,
      3. Neilands TB,
      4. et al
      . The effect of executive walk rounds on nurse safety climate attitudes: a randomized trial of clinical units. BMC Health Serv Res 2005;5:28.
      OpenUrlCrossRefPubMed
    41. ↵
      1. Krauss MJ,
      2. Evanoff B,
      3. Hitcho E,
      4. et al
      . A case-control study of patient, medication, and care-related risk factors for inpatient falls. J Gen Intern Med 2005;20:116–22.
      OpenUrlCrossRefPubMedWeb of Science
    42. ↵
      1. Gelinas L,
      2. Bohlen C
      . The business case for retention. J Clin Syst Manage 2002;4:14–16.
      OpenUrl
    43. ↵
      1. Jones CB
      . The costs of nurse turnover, part 2: application of the Nursing Turnover Cost Calculation Methodology. J Nurs Adm 2005;35:41–9.
      OpenUrlPubMedWeb of Science
    44. ↵
      1. Needleman J,
      2. Buerhaus P,
      3. Mattke S,
      4. et al
      . Nurse-staffing levels and the quality of care in hospitals. N Engl J Med 2002;346:1715–22.
      OpenUrlCrossRefPubMedWeb of Science
      1. O'Brien-Pallas L,
      2. Griffin P,
      3. Shamian J,
      4. et al
      . The impact of nurse turnover on patient, nurse, and system outcomes: a pilot study and focus for a multicenter international study. Policy Polit Nurs Pract 2006;7:169–79.
      OpenUrlAbstract/FREE Full Text
    46. ↵
      1. Hayes LJ,
      2. O'Brien-Pallas L,
      3. Duffield C,
      4. et al
      . Nurse turnover: a literature review. Int J Nurs Stud 2006;43:237–63.
      OpenUrlCrossRefPubMedWeb of Science
    47. ↵
      1. Bae SH,
      2. Mark B,
      3. Fried B
      . Impact of nursing unit turnover on patient outcomes in hospitals. J Nurs Scholarsh 2010;42:40–9.
      OpenUrlCrossRefPubMed
    48. ↵
      1. Mark BA,
      2. Salyer J,
      3. Wan TT
      . Professional nursing practice: impact on organizational and patient outcomes. J Nurs Adm 2003;33:224–34.
      OpenUrlCrossRefPubMedWeb of Science
    49. ↵
      1. Hart SE
      . Hospital ethical climates and registered nurses' turnover intentions. J Nurs Scholarsh 2005;37:173–7.
      OpenUrlCrossRefPubMedWeb of Science
    50. ↵
      1. Stordeur S,
      2. D'Hoore W
      . Organizational configuration of hospitals succeeding in attracting and retaining nurses. J Adv Nurs 2007;57:45–58.
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      1. Breslin FC,
      2. Smith P
      . Trial by fire: a multivariate examination of the relation between job tenure and work injuries. Occup Environ Med 2006;63:27–32.
      OpenUrlAbstract/FREE Full Text
    52. ↵
      1. Kim H,
      2. Kriebel D,
      3. Quinn MM,
      4. et al
      . The snowman: a model of injuries and near misses for the prevention of sharps injuries. Am J Ind Med 2010;53:1119–27.
      OpenUrlCrossRefPubMed
    53. ↵
      1. Blegen MA,
      2. Vaughn T,
      3. Pepper G,
      4. et al
      . Patient and staff safety: voluntary reporting. Am J Med Qual 2004;19:67–74.
      OpenUrlAbstract/FREE Full Text
    54. ↵
      1. Siddharthan K,
      2. Hodgson M,
      3. Rosenberg D,
      4. et al
      . Under-reporting of work-related musculoskeletal disorders in the Veterans Administration. Int J Health Care Qual Assur Inc Leadersh Health Serv 2006;19:463–76.
      OpenUrlPubMed
    55. ↵
      1. Lake ET,
      2. Shang J,
      3. Klaus S,
      4. et al
      . Patient falls: association with hospital Magnet status and nursing unit staffing. Res Nurs Health 2010;33:413–25.
      OpenUrlCrossRefPubMedWeb of Science
    56. ↵
      1. Dunton N,
      2. Gajewski B,
      3. Taunton RL,
      4. et al
      . Nurse staffing and patient falls on acute care hospital units. Nursing Outlook 2004;52:53–9.
      OpenUrlCrossRefPubMedWeb of Science
    57. ↵
      1. Haberfelde M,
      2. Bedecarr D,
      3. Buffum M
      . Nurse-sensitive patient outcomes: an annotated bibliography. J Nurs Adm 2005;35:293–9.
      OpenUrlPubMed
    58. ↵
      1. Unruh L
      . Licensed nurse staffing and adverse events in hospitals. Med Care 2003;41:142–52.
      OpenUrlCrossRefPubMedWeb of Science
    59. ↵
      1. Colla JB,
      2. Bracken AC,
      3. Kinney LM,
      4. et al
      . Measuring patient safety climate: a review of surveys. Qual Saf Health Care 2005;14:364–6.
      OpenUrlAbstract/FREE Full Text
    60. ↵
      1. Sexton J,
      2. Thomas E,
      3. Pronovost P
      . The context of Care and the Patient Care Team: The Safety Attitudes Questionnaire. In: Reid PP, Compton WD, Grossman JH, Fanjiang G, eds. Building a Better Delivery System. A New Engineering Health Care Partnership. Washington, DC: National Academies Press, 2005:119–23.

    Supplementary materials

    Footnotes

    • Funding This research was supported (in part) by funding from the NIOSH Education and Research Center for Occupational Safety and Health at the Johns Hopkins Bloomberg School of Public Health—a doctoral training program (#T42OH00842428). Dr Taylor was a trainee in this program from 2003 to 2007.

    • Competing interests None.

    • Ethics approval Johns Hopkins Bloomberg School of Public Health.

    • Provenance and peer review Not commissioned; externally peer reviewed.